Professional Documents
Culture Documents
Why did the United States subsidize American multinationals’ entry into countries
treated as informal colonies? We study a classic case of American imperialism,
the 1903 U.S. support of Panama’s secession from Colombia and subsequent U.S.
SD\PHQWRI WKH UHSDUDWLRQVWKDW RSHQHG &RORPELD¶VRLO ¿HOGVWR6WDQGDUG
Oil. We test Noel Maurer’s (2013) empire trap hypothesis quantitatively. Archival
and econometric evidence documents Colombia’s threat to Standard Oil’s sunk
investment, which induced the multinational to build a supermajority coalition
in the U.S. Senate to back a reparations treaty. Results support the empire trap
K\SRWKHVLVEXWSRLQWRXWLPSRUWDQWTXDOL¿FDWLRQV
The Journal of Economic History, Vol. 77, No. 1 (March 2017). © The Economic History
Association. All rights reserved. doi: 10.1017/S0022050717000055
Xavier Duran is Associate Profesor, Universidad de los Andes Management School, Calle
21 No. 1-20, Bogota, Colombia. E-mail: xh.duran21@uniandes.edu.co. Marcelo Bucheli is
Associate Professor, University of Illinois at Urbana-Champaign, 198 Wohlers Hall, Champaign,
IL 61820. E-mail: mbucheli@illinois.edu.
We wish to thank Paul Rhode, the referees, Gareth Austin, Ed Balleisen, Raquel Bernal,
Agustin Casas, Charles Calomiris, Luis Castañeda, Paulo Drinot, Alan Dye, Marcela Eslava,
Nadia Fernández de Pinedo, Paloma Fernández, Price Fishback, Leopoldo Fergusson, Leigh
Gardner, Andrew Godley, Alejandra Irigoin, Andrea Lluch, Luis Fernando Medina, Tomas
Nonnenmacher, Monica Pachón, Marc Prat, Juan Carlos Rodríguez, Alan Rugman, Patricio Saíz,
Fabio Sanchez, Rodrigo Taborda, Randall Walsh, Benjamin Waterhouse, John Wallis, Susan
Wolcott, and participants of the Cliometric Society Meeting (2016), Business History Conference
(2014), World Business History Congress (2014), Economic History Association (2014),
Economic History Society (2014), and workshops at the University of London (2013), University
of Reading (2013), Universidad Autónoma de Madrid (2014), Universidad de Barcelona (2014),
and Universidad de los Andes (2015) for their comments. We are grateful to Price Fishback who
generously shared federal tax and expenditure, and personal income data with us. Luis Felipe
Sáenz and Julián Gómez provided great research assistance.
251
https://doi.org/10.1017/S0022050717000055 Published online by Cambridge University Press
252 Duran and Bucheli
social issues and faced low political competition (and whose votes were
plausibly bought); and senators who leaned consistently to prefer govern-
ment intervention on social issues. Opposition came from senators repre-
senting states that were net federal tax contributors and from senators
who consistently opposed government intervention on social issues and
were associated with President Theodore Roosevelt. Having pushed for
Panama’s secession, the former president objected to reparations.
7KH¿QGLQJVFRQWULEXWHWRVHYHUDOOLWHUDWXUHV)LUVWHPSLULFDOWUDGHDQG
economic history literature suggest two rationales for imperial subsidies
to its multinationals: the empire effect and the empire trap.
In the empire effect rationale, an empire may offer subsidies in the form
of common law, language, currency, or lower trade barriers to promote
a positive externality and increase welfare (Mitchener and Weidenmier
2005, 2008; Ferguson and Schularick 2006).
In the empire trap argument, that is more closely connected to our study,
a multinational entering a formal or informal colony faces opportunistic
behavior and resistance from the colony, and in an informal colony it also
faces a risk of expropriation. An empire’s multinational exposed to these
risks has incentives to lobby the imperial government for a subsidy that
UHGXFHVWKH¿UP¶VH[SRVXUH7KHVXEVLG\PD\EHDGLUHFWSD\PHQWWRWKH
multinational, but more frequently has been aid to the colony in exchange
for protection for the multinational assets, or a threat of military action
(Hopkins 1973; Frankema 2010; Maurer 2011). Extending the logic of
this argument, Noel Maurer (2013) has suggested that an empire is likely
to undertake the requested intervention even if the capital exposed to
risk is small and no positive externality for the empire exists. Failure
of political collective action a la Mancur Olson (1965) may exist if the
PXOWLQDWLRQDOLVZHOORUJDQL]HGDQGH[SHFWVWRFRQFHQWUDWHWKHEHQH¿WVRI
intervention, while costs are expected by decentralized and unorganized
groups of individuals who each face a small loss. Thus, a multinational
is likely to gain subsidies even if it is not in the interest of the empire
as a whole—the empire trap. A series of American imperial interven-
tions to protect the property of their multinationals has been documented
DQGLPSOLFLWVXEVLGLHVLQFUHDVHGWKHEHQH¿WHG¿UPV¶FDSLWDOPDUNHWYDOXH
(Dube, Kaplan, and Naidu 2011; Maurer 2013).1
$V IDU DV ZH DUH DZDUH ZH SURYLGH WKH ¿UVW TXDQWLWDWLYH WHVW RI WKH
empire trap hypothesis. The test reveals that collective action failure in
WKHSROLWLFDOPDUNHWGRHVSOD\DUROH2LOVSHFLDOLQWHUHVWVLQÀXHQFHGVRPH
1
/HEHUJRWW DUJXHV86LQWHUYHQWLRQVJHQHUDWHGVPDOOEHQH¿WVWR86SULYDWHGRPHVWLF
and foreign investment, 1890–1929.
senators and plausibly bought the votes of others to ratify the Colombian
reparations treaty. But senators generally favoring government interven-
tion in social issues also played an important role. Ideology, as much as
economics, is an important driver of the political process of the empire
WUDSLQWKH8QLWHG6WDWHV7KHLPSRUWDQFHRIWKHUHGLVWULEXWLYHFRQÀLFWLV
DOVRTXDOL¿HG$OWKRXJKWRWDOEHQH¿WVIRUWKH8QLWHG6WDWHVZHUHKLJKHU
WKDQWKHUHSDUDWLRQSDLGWR&RORPELDWKHUH¿QHUVFDSWXUHGPRVWRIWKHVH
EHQH¿WVZKLOHWKH$PHULFDQRLOGHULYDWLYHVFRQVXPHUVDQGLQGLYLGXDOV
DQGQRQRLO¿UPVWD[SD\HUVHQGHGXSOHVVZHOORIIWKDQEHIRUHWKHWUHDW\
UDWL¿FDWLRQ
Second, economic and business history and the international business
OLWHUDWXUHKDYHTXDOL¿HGWKHLGHDWKDWDPXOWLQDWLRQDOZLOOEHVXSSRUWHGE\
its home empire and that the empire may exercise unchallenged power
over an informal colony and so determine outcomes in these countries.
Particularly in informal empires, the governments of colonies can hold
up multinational companies and leverage negotiated outcomes (Kobrin
1980; Maurer 2011, 2013). Our results qualify further the idea of unchal-
lenged exercise of imperial power and the role of multinationals in the
HPSLUHFRORQ\ JRYHUQPHQW UHODWLRQVKLS :H ¿QG WKDW D VSHFLDO LQWHUHVW
JURXS FDQ LQÀXHQFH LWV RZQ HPSLUH¶V SROLF\ LQ IDYRU RI DQ LQIRUPDO
colony, induced by the colony’s government via hold up pressure on the
special interest group’s multinational local operations. As far as we are
DZDUHQHLWKHUWKHSRLQWWKDWDQLQIRUPDOFRORQ\¶VJRYHUQPHQWPD\LQÀX-
ence an empire’s policy nor the multinational as a mechanism to do so
KDVEHHQLGHQWL¿HGEHIRUHLQWKHOLWHUDWXUH
Third, rather than studying either negative imperial interventions like
military interventions or coups (Dube, Kaplan, and Naidu 2011; Berger
et al. 2013) or positive ones like aid or reparations (Ball and Johnson
1996; Alesina and Dollar 2000), we show that these two types of impe-
rial interventions are frequently connected and should be studied as part
of a wider and more dynamic bargaining process between the informal
colony and empire governments.
2Q$SULOWKH866HQDWHUDWL¿HGWKH8UUXWLD7KRPVRQ7UHDW\
(UTT). The UTT was a foreign treaty that committed the United States to
pay $25 million in reparations to Colombia in compensation for American
support of Panama’s secession from Colombia in 1903. In this section we
H[DPLQHWKHHYHQWVOHDGLQJWRUDWL¿FDWLRQRIWKHWUHDW\DQGGRFXPHQWWKH
2
Colombia, Ministerio de Minas y Petróleo (1929, no. 8, pp. 90–92); New York Times, 7
January 1920, 20 January 1920, 14 August 1920, 23 August 1920, 8 October 1920; Wall Street
Journal, 23 August 1920; Bell (1921, pp. 120, 128); Gibb and Knowlton (1956, pp. 85, 108, 109).
%XW621-IDFHGDQRWKHUKXUGOH&RORPELD¶VRLO¿HOGZDVORFDWHGLQWKH
center of the country, 300 miles from the Caribbean coast where crude
FRXOGEHORDGHGRQWRWDQNHUVDQGVHQWWR621-UH¿QHULHVLQ1HZ-HUVH\
or Canada. Thus, SONJ asked the Colombian government for a conces-
sion to build a 300-mile oil pipeline.
7KH866HQDWH5DWL¿FDWLRQRIWKH8UUXWLD7KRPVRQ7UHDW\
3
New York Times, 20 February 1917, 19 June 1919, 26, 27 July 1919. Coatsworth (2006, p.
0DXUHUDQG<X S 0XUSK\ S 3DODFLRVDQG6DIIRUG S
4
Colombia, Cámara de Representantes (1925, p. 18).
1921, Senator James Reed, D-MO, clearly stated the case put forward by
the oil interests by observing that “… an attorney for these oil compa-
QLHV«FDPHWR:DVKLQJWRQDQGVWDWHGWKDWLIWKHWUHDW\ZDVQRWUDWL¿HG
it would involve the entire oil situation; that the present administration
[in Colombia] might be overthrown and that the oil interests of these
[American] people lost. The substance of the talk was that the treaty must
EHUDWL¿HGLQRUGHUWRSURWHFWWKH>$PHULFDQ@RLOLQWHUHVWV´ &RQJUHVVLRQDO
Record, 67th Congress, 1st Session, p. 314). With the death of former
President Theodore Roosevelt, some Old Guard Republicans decided to
change their position. Senator Lodge, a member of the Foreign Relations
&RPPLWWHH VLQFH DQG FRQVLVWHQWO\ RSSRVHG WR UDWL¿FDWLRQ RI WKH
WUHDW\ QRZ LQGLFDWHG WKDW ³WKH UDWL¿FDWLRQ RI WKLV 7UHDW\ ZLOO OHDG WR D
prompt additional treaty of amity and commerce with Colombia [presum-
ably the oil pipeline concession] which will improve our opportunities
there making secure the concessions we now have.” He read a letter sent
by Secretary of the Interior Fall, a former senator and member of the
Foreign Relations Committee and consistently opposed to the treaty, now
writing “I have every assurance … short of actual written agreement that
the present Colombian government and prominent Colombians, favoring
WKLVSROLF\ZLOOLPPHGLDWHO\XSRQUDWL¿FDWLRQRIWKHSUHVHQWWUHDW\«
enter into a supplemental treaty [presumably the pipeline concession]
….” (Congressional Record, 67th Congress, 1st Session, pp. 116, 163).
And Senator Porter McCumber, R-ND, another former opponent of the
treaty now stated, “I am voting to stake $25m on the effort of the presi-
dent to secure without an additional donation a supplemental agreement
that will be worth to this country many times that sum” (Congressional
Record, 67th Congress, 1st Session, p. 116).
Progressive Republican senators opposed the UTT. Senator Hiram
Johnson from California, who had been running mate of Theodore
Roosevelt for the Progressives during the 1912 election, asked in his
congressional speech “Why do we have $25,000,000 to squander
LQ WKH ¿UVW DFW WKDW D 5HSXEOLFDQ DGPLQLVWUDWLRQ GRHV´ DQG GHQRXQFHG
Republicans who had changed their mind asking “tell me when the
blackmail demand shed its awful outer garment and became a rosy-hued
request” (Murphy 2013, p. 569). Senator William Kenyon from Iowa
characterized Colombia’s pressure as blackmail and observed the political
and economic opportunity cost the treaty implied when he indicated, “I
wonder what these gentlemen who have raised their voice so loudly about
the [advantages of the UTT for the United States] economy are going to
say when the soldier bonus bill comes here, and they have the record
of voting away $25 million in a blackmail proposition” (Congressional
Record, 67th Congress, 1st Session, p. 472). Senator George Norris from
Nebraska, accepted that Roosevelt and the American government were
at fault during the secession of Panama, but thought, “let the oil, rather
than the Treasury of the United States, pay for the smiles we are trying
to get” (Murphy 2013, p. 569). Senator William Borah from Idaho, after
confessing to Arthur H. Vandenberg, editor of the Grand Rapids Herald,
that “We must strike fast and strike hard for the lobbying behind this
thing is simply stupendous,” prepared a roll call vote amending the treaty
by absolving the United States for aiding Panama’s secession, but this
was defeated (Murphy 2013, p. 569). On 20 April 1921 the UTT was
UDWL¿HG
)ROORZLQJ UDWL¿FDWLRQ &RORPELD DZDUGHG WKH SLSHOLQH FRQFHVVLRQ
to the Andian Corporation in 1923.5 Andian’s ownership was trans-
ferred to SONJ’s subsidiary International Petroleum Company in 1925,
centralizing production and transportation operations. The pipeline was
FRPSOHWHGLQ0DUFKDQGWKH¿UVWWDQNHUVHWRIIWRWKH8QLWHG6WDWHV
on 3 July 1926. By 1928, Colombia was the world’s eighth largest oil
producer.6 Between then and the end of SONJ’s concession in 1951, the
multinational controlled almost the totality of Colombian oil exports and
production (Bucheli 2008, p. 80). Colombia could not change the terms
RIWKHFRQWUDFWZLWK621-DIWHUWKH877UDWL¿FDWLRQ7KHPLOOLRQZDV
paid between 1923 and 1926, most of it after the 1923 Andian concession
contract was awarded.7 This assured that the deal the U.S. Senate voted
for was completed before most money was paid.
7+(85587,$7+2062175($7<927($1'7+(2,/,17(5(676
5
Colombia, Cámara de Representantes (1925, pp. 27, 41), Colombia, Ministerio de Minas y
Petróleo (1929, vol. 2, no. 8, pp. 99–110), Gibb and Knowlton (1956, pp. 378–80), Rippy (1976,
p. 121), de la Pedraja (1985, pp. 191–93).
6
Oil and Gas Journal, 22 January 1925, p. 22, 5 February, p. 120, 7 May, p. 68; Gibb and
Knowlton (1956, pp. 379–80, p. 659); Petroleum Facts and Figures, 1929, p. 4
7
Colombia, Ministerio de Hacienda, Memoria de Hacienda (1926).
plenary vote. One round of plenary debate, a roll call vote, and two-thirds
positive votes are required to pass the proposal (U.S. Senate Committee
of Foreign Relations 2000, p. 20).
The 20 April 1921 UTT roll call was voted on by 88 senators, which
required at least 59 senators to vote positively to pass the treaty, and in
fact 69 voted to ratify it. Democrats, the Senate minority, supported rati-
¿FDWLRQE\SHUFHQWZKLOH5HSXEOLFDQPDMRULW\E\SHUFHQW
:DVWKLVELSDUWLVDQPDMRULW\LQÀXHQFHGE\621-DQGWKHRLOLQWHUHVWV"
6HQDWRUVPRVWOLNHO\WREHLQÀXHQFHGZHUHIURPVWDWHVZKHUH621-KDG
operations. SONJ operated in ten states, and all but two of those senators
(who abstained) supported the treaty, 18 of the 59 votes necessary to pass
the treaty.8
7KH LQWHUHVW RI RWKHU RLO UH¿QHUV ZDV DOLJQHG ZLWK WKDW RI 621-
Colombia’s entry into the U.S. crude oil market was expected to make
the supply more elastic, pressing equilibrium prices down. Since crude
RLOUHSUHVHQWHGDERXWSHUFHQWRIWKHDYHUDJHXQLWFRVWRIUH¿QHGSURG-
XFWVDQGWKHLQGXVWU\ZDVFRPSRVHGE\UHJLRQDOROLJRSROLHVRLOUH¿QHUV
would reap substantial gains from crude oil price reductions and would
QRWSDVVWKHVHWR¿QDOFRQVXPHU 2OPVWHDGDQG5KRGH+RSNLQV
1927; U.S. Census 1921, Vol. Manufactures, p. 762).
,QSULQFLSOHWKHFRVWVWUXFWXUHDQGSULFLQJEHKDYLRURIUH¿QHUVLPSO\
WKH\ZRXOGEHQH¿WIURPWKHWUHDW\ZKLOHFUXGHRLOSURGXFHUVZRXOGORVH
via a drop in the output price. However, if there is a larger absolute level
RISUR¿WWREHPDGHLQUH¿QLQJFRPSDUHGWRWKHORVVLQFXUUHGLQSURGXF-
WLRQ DQG LI WKH ¿UPV SURGXFLQJ FUXGH RLO DQG RLO GHULYDWLYHV DUH LQWH-
JUDWHGLWLVSRVVLEOHWKDWWKHFRQÀLFWEHWZHHQFUXGHRLOSURGXFHUVDQGRLO
UH¿QHUVPD\LQIDFWEHLQWHUQDOL]HGWRDODUJHH[WHQWZLWKLQ¿UPV,QWKLV
case the entire industry would support the treaty.
Although we are not aware of contemporary systematic data on
YHUWLFDO LQWHJUDWLRQ LQ WKH RLO LQGXVWU\ DFURVV VWDWHV VLJQL¿FDQW H[DP-
ples do abound. For instance, most of the “baby standards” companies
the Supreme Court of the United States created in the 1911 break-up
of Standard Oil Holding Company are important and relevant examples
of vertical integration (Hidy and Hidy 1955). The major oil companies
LQWHJUDWHGSURGXFWLRQDQGUH¿QLQJ,QPDMRUVLQWHJUDWHGDERXW
SHUFHQW RI 86 SURGXFWLRQ DQG SHUFHQW RI UH¿QHG SURGXFWV 86
Congress 1941, p. 51).
The U.S. Census (1921, Vol. Mines, 28, Vol. Manufactures, p. 758)
UHSRUWV WKDW VWDWHV SURGXFHG FUXGH RLO UH¿QHG RLO DQG D WRWDO RI
8
The two abstaining senators were from Oklahoma and West Virginia.
SURGXFHGRUUH¿QHGRLO$VVXPLQJWKHRLOFRPSDQLHVZHUHVXFFHVVIXO
LQFRQYLQFLQJWKHLURZQVHQDWRUVWRYRWHIRUUDWL¿FDWLRQ\LHOGVDWRWDORI
58 votes, one fewer than the minimum two-thirds necessary to ratify the
treaty. In fact, 45 of the 53 senators from these states who participated in
the UTT roll call did vote to support the treaty. A vertically integrated oil
LQWHUHVWLQÀXHQFHGPDQ\VHQDWRUVEXWWKLVLVVWLOOQRWHQRXJKWRH[SODLQ
UDWL¿FDWLRQ
:+$7(/6(,1)/8(1&('6(1$7256725$7,)<7+(75($7<"
9
The 67th Congress at the time of the UTT roll call vote had 96 senators. One each from
Florida, Iowa, Michigan, Montana, Oklahoma, Tennessee, Vermont, and West Virginia preferred
to abstain.
10
Estimated marginal effects for CIs are qualitatively identical to those estimated by using the
ratio of state federal taxes over expenditure. Data was kindly provided by Price Fishback.
PRUHLQÀXHQFHRLOSURGXFHUVKDYHRQKLVYRWHUHODWLYHWRKLVLGHRORJ\
and constituency interests. Thus, the share of crude oil production in the
value of overall production in state s can be a proxy for the oil producers’
special interests contributions, SIP.
5H¿QHUVH[SHFWHGWRJDLQIURPWKH877VLQFHWKHSULFHRIWKHLUNH\
input was expected to fall. Following a logic analogous to that approxi-
mating SIPUH¿QHUV¶VSHFLDOLQWHUHVWFRQWULEXWLRQV SIR, use the share of
FUXGHRLOFRQVXPSWLRQLQUH¿QLQJLQVWDWHs as a proxy. Data for crude
RLOSURGXFHUDQGUH¿QHUVSHFLDOLQWHUHVWVZDVGUDZQIURPWKH86
FHQVXV 7KH GH¿QLWLRQ DQG VRXUFHV RI DOO YDULDEOHV XVHG LQ WKH HFRQR-
metric analysis and its descriptive statistics are included in Appendix
Tables 1 and 2.
7KH,QÀXHQFHRI2LO,QWHUHVWVRQ9RWLQJ3DWWHUQV
TABLE 1
INFLUENCE OF OIL INTERESTS ON UTT VOTING PATTERNS
WKHVHWZRPRGHOVFRQ¿UPWKHLQWXLWLRQWKDWYHUWLFDOLQWHJUDWLRQPD\KDYH
KHOSHGWRDOLJQWKHLQWHUHVWVRIRLOSURGXFHUVDQGUH¿QHUV7KHHVWLPDWHRI
DYHUWLFDOO\LQWHJUDWHGRLOLQWHUHVWLQÀXHQFHLQVSHFL¿FDWLRQLVSRVLWLYH
DQGVWDWLVWLFDOO\VLJQL¿FDQWZKLOHWKHHVWLPDWHVIRUWKHYHUWLFDOO\GLVLQWH-
JUDWHGRLOYDOXHFKDLQLQVSHFL¿FDWLRQLQFROXPQDUHQRWVLJQL¿FDQW7KH
HVWLPDWHRIWKHLQÀXHQFHRIVHQDWRUV¶SRVLWLRQRQJRYHUQPHQWHFRQRPLF
intervention (W1 RQYRWLQJSDWWHUQVLVQRWVLJQL¿FDQWZKLOHWKHHVWLPDWH
IRUWKHLQÀXHQFHRIKLVSRVLWLRQRQJRYHUQPHQWVRFLDOLQWHUYHQWLRQ W2)
LVSRVLWLYHKLJKO\VWDWLVWLFDOO\VLJQL¿FDQWDQGLQFUHDVHVWKHDGMXVWHG5
VXEVWDQWLDOO\7KHHVWLPDWHRIWKHLQÀXHQFHRIFRQVWLWXHQF\LQWHUHVWV WD[
UDWLR LVDVH[SHFWHGQHJDWLYHDQGVLJQL¿FDQWDWSHUFHQW
,QVSHFL¿FDWLRQLQFROXPQDIWHULQFOXVLRQRIWKHVHQDWRUV¶SRVLWLRQRQ
government economic and social intervention and constituency interests,
WKHHVWLPDWHRIYHUWLFDOO\LQWHJUDWHGRLOLQWHUHVWLQÀXHQFHLQFUHDVHVIURP
1.19 to 1.52. A one standard deviation increase in vertically integrated
oil interests leads to an increase of 7.4 percent in the likelihood a given
senator voted to ratify the treaty, while a one standard deviation increase
favoring government intervention in social issues leads to a 22.7 percent
LQFUHDVHLQWKHOLNHOLKRRGWKHVHQDWRUUDWL¿HGWKHWUHDW\$QLQFUHDVHLQWKH
tax ratio so that a state moves from being the 22nd net contributor to being
the 10th highest net contributor (equivalent to a one standard deviation
increase) leads to a 10 percent decline in the likelihood a given senator
UDWL¿HGWKHWUHDW\
)LQDOO\ VSHFL¿FDWLRQV LQ FROXPQV ± LQFOXGH VWDWH FRQWUROV 7KH
PRGHOVFRQ¿UPWKDWSURGXFHUVDQGUH¿QHUVGLGQRWH[HUWRSSRVLQJLQÀX-
ence on the senators; more likely vertically integrated oil interests within
HDFKVWDWHLQÀXHQFHGWKHVHQDWRUV¶YRWHV7KHHIIHFWRIWKHLQÀXHQFHRI
the vertically integrated oil interest is robust to inclusion of state census
FRQWUROVDQGVWDWH¿[HGHIIHFWV7KHHIIHFWRIWKHLQÀXHQFHRIVHQDWRUV¶
position on government social intervention on voting patterns is also
UREXVWWRLQFOXVLRQRIVWDWHFRQWUROV7KHLQÀXHQFHRIFRQVWLWXHQF\LQWHU-
HVWVLVUREXVWWRVWDWHFHQVXVFRQWUROVEXWQRWWRVWDWH¿[HGHIIHFWVSRVVLEO\
a consequence of collinearity.11
9RWH%X\LQJDQGWKH8UUXWLD7KRPVRQ7UHDW\9RWH2XWFRPH
7KHEDVHOLQHVSHFL¿FDWLRQLQ7DEOHUHSUHVHQWVDVLPSOHPRGHO$Q
LPSRUWDQWLQÀXHQFHRQVHQDWRULDOYRWLQJSDWWHUQVWKDWDOVRGHVHUYHVDWWHQ-
tion is plausible vote buying. James Snyder (1991) suggests that a special
11
Unreported logit model results are qualitatively identical to the linear probability model
results.
interest group will focus on buying the votes of legislators who are
slightly opposed to the proposed law, rather than the votes of legislators
who strongly support or oppose the law.
Recall that for the W-Nominate score the lower and higher score levels
UHÀHFWVHQDWRULDOFRQVLVWHQWFKRLFHDJDLQVWRUIRUJRYHUQPHQWLQWHUYHQ-
tion. The range of scores close to 0, the center point on each dimen-
sion, indicate a senator vote is against and for government interven-
tion roughly as frequently. Senators may behave this way for different
reasons. For instance, legislators may be truly indifferent to government
intervention and decide on a case by case basis; or senators may prioritize
their re-election probability in their utility function, and sell their vote
and follow different special interest groups and different positions on
different votes. However, even if a W-Nominate score close to 0 does
QRWDOORZLQIHUULQJGLUHFWO\WKHLGHRORJLFDORUQRQLGHRORJLFDOLQÀXHQFHV
over a senator’s voting pattern, the score does help to identify senators
who swing from one position in one roll call vote to another position
in another vote. Following Snyder’s conclusion, senators with W score
close to 0 represent targets for special interest groups to induce them to
sell their vote.
We use the typical position of a senator on government intervention
in social issues and construct a dummy variable to identify senators who
showed no consistent voting patterns. Since the W2 score ranges from
±WRWKHGXPP\YDULDEOHLGHQWL¿HVVHQDWRUVZKRVHDEVROXWHYDOXHRI
the W2i score is within the top 50th percentile closest to 0 (Indifferent),
the central point of the W2 dimension, indicating senator i choice was
not consistent, was against and for government intervention roughly as
frequently. Senators with W2i VFRUH GH¿QHG DV LQGLIIHUHQW SUHVXPDEO\
face lower utility losses from changing their position, signal their will-
ingness to change their position more frequently, and their votes will be
in high demand by special interest groups.
Additionally, we use electoral competition data to construct an alter-
native indicator of plausible vote buying that does not rely on voting
behavior during the 67th Congress. The vote percentage margin between
the elected senator and the second runner up is measured by the variable
Margin. Senators facing a higher margin and lower electoral competition
¿QGLWOHVVFRVWO\WRGHYLDWHIURPWKHLUFRQVWLWXHQF\LQWHUHVWVWKHLUYRWHV
are cheaper to buy and will be in high demand.
,Q 7DEOH WKH EDVHOLQH PRGHO LQ 7DEOH VSHFL¿FDWLRQ LQ FROXPQ
LV H[SDQGHG WR LQFOXGH WKH HIIHFWV RI YRWH EX\LQJ 7KH ¿UVW UHVXOW WR
KLJKOLJKWLVWKDWDFURVVWKHWKUHH¿UVWVSHFL¿FDWLRQVLQ7DEOHWKHEDVH-
line model estimates for oil interests, senators’ position on government
TABLE 2
,1)/8(1&(2)3/$86,%/(927(%8<,1*21877927,1*3$77(516
6SHFL¿FDWLRQ (1) (2) (3) (4) (5)
Peace with
Dependent Variable Ratify Ratify Ratify Reduce Germany
negative; the F test indicates the joint hypothesis that Oil interests, W2,
and Indifferent are equal to 0 can be rejected. Thus, the coalition that
UDWL¿HGWKH7UHDW\ZDVDVWURQJRQHQRWHDV\WREUHDNXSDQGDFWHGWR
oppose the alternative Treaty text. It preferred to favor Colombia than
American taxpayers.
Second, in a placebo test, when we use the same set of independent
variables that explain the UTT vote to explain an entirely different Senate
YRWHWKHHVWLPDWHVRIWKHVHYDULDEOHVVKRXOGQRWEHVLJQL¿FDQW9DULDEOHV
QRWVLJQL¿FDQWLQWKHEDVHOLQHPRGHOPD\EHVLJQL¿FDQWDQGFRQWULEXWH
importantly to explanatory power. This test can help to rule out the possi-
ELOLW\WKDWWKHFRDOLWLRQWKDWUDWL¿HGWKH877ZDVLQIDFWDQH[WHQVLRQRID
broader coalition based over a larger set of issues and roll call votes during
the 67th Congress. Less than 10 percent of all roll calls voted during the
67th Congress were performed by at least the same 88 senators and six of
these were foreign treaties. Column 5 presents results for the roll call vote
to sign peace with Germany. The individual estimates of oil interest, W2,
and IndifferentDUHQRWVLJQL¿FDQWDQGWKH)WHVWVIRUZKHWKHUWKHVHWKUHH
HVWLPDWHVDUHGLIIHUHQWIURPLVDOVRQRWVLJQL¿FDQW7KHWKUHHYDULDEOHV
representing the coalition that passed the UTT treaty were unlikely to
LQÀXHQFHWKHYRWHRQSHDFHZLWK*HUPDQ\6LPLODUUHVXOWVDUHREVHUYHG
IRUWKHRWKHU¿YHIRUHLJQWUHDWLHV7KXVLWLVXQOLNHO\WKHFRDOLWLRQWKDW
passed the UTT represented a broad coalition over the 67th Congress.
Table 3 summarizes the econometric evidence on the oil coalition.
Lines 1 and 2 indicate that 88 senators participated in the roll call vote
DQGYRWHGWRUDWLI\WKH8777KH¿JXUHVLQOLQHVWRRIWKHWDEOH
DUHWKHQXPEHURISUHGLFWHGYRWHVVXSSRUWLQJUDWL¿FDWLRQDVVXPLQJWKDW
when the probability that senator i votes yes is 0.5 or higher he should
KDYHYRWHGWRUDWLI\²WKHVWDQGDUGDVVXPSWLRQLQFRQVWUXFWLQJFODVVL¿FD-
tion tables for binary data models. Line 3 indicates the model in Table 2
VSHFL¿FDWLRQLQFROXPQSUHGLFWVWKDWVHQDWRUVVKRXOGKDYHYRWHGWR
ratify the UTT, on the basis of their position in W1 and W2 alone—the
HVWLPDWHGFRHI¿FLHQWRQW1 times the score of senator i on W1 plus the
HVWLPDWHGFRHI¿FLHQWRQW2 times the score of senator i on W2 is higher
WKDQIRUVHQDWRUV)RUVSHFL¿FDWLRQLQFROXPQLWLVVHQDWRUV
Thus, if the UTT decision had been based solely on the senators positions
in W1 and W2WKHPRGHOVVXJJHVWLWVKRXOGKDYHEHHQUDWL¿HGEHFDXVH
the predicted positive votes were higher than 59.
/LQH LQGLFDWHV WKH VLPXOWDQHRXV LQÀXHQFH RI W1 and W2, and tax
FRQVLGHUDWLRQV RQ YRWLQJ SDWWHUQV 0RGHO VSHFL¿FDWLRQ LQ FROXPQ
predicts that 55 senators should have voted to ratify, while column 2,
also including the negative effect of the Margin estimate, predicts 58.
TABLE 3
180%(52)6(1$72567+$7927('$1'35(',&7('725$7,)<877
0RGHOVSHFL¿FDWLRQLQ7DEOH (1) (3)
Actual votes
(1) Senators that voted UTT 88 88
(2) Senators that voted UTT yes 69 69
Predictions
(3) W1+W2 predicts yes 70 73
(4) W1+W2 + Tax ratio + Margin predicts yes 55 58
::7D[UDWLR2LOLQÀXHQFHSUHGLFWV\HV 68 69
(6) W1+W2 predicts no 18 15
::2LOLQÀXHQFHSUHGLFWV\HV JLYHQ::SUHGLFWVQR 3 3
Note 0RGHO VSHFL¿FDWLRQ DQG FRUUHVSRQG WR PRGHO VSHFL¿FDWLRQ DQG LQ 7DEOH
UHVSHFWLYHO\ 0RGHOV ZHUH UHHVWLPDWHG XVLQJ UHVFDOHG YDULDEOH YDOXHV WR ¿W LQWHUYDO DQG
produce predicted values in 0-1 interval. W1 + W2 predicts yes is the number of senators for
ZKLFK WKH VXP RI SUHGLFWHG SUREDELOLW\ FRHI¿FLHQW HVWLPDWH : W1i FRHI¿FLHQW HVWLPDWH
W2*W2i) is higher than 0.5. W1 + W2 + Tax ratio predicts yes is the number of senators for which
WKHVXPRIWKHSUHGLFWHGSUREDELOLW\ FRHI¿FLHQWHVWLPDWH: W1iFRHI¿FLHQWHVWLPDWH: W2i
FRHI¿FLHQWHVWLPDWHWD[UDWLR tax ratioiFRHI¿FLHQWHVWLPDWHPDUJLQ Margini) is higher than
::7D[2LOLQÀXHQFHLVWKHQXPEHURIVHQDWRUVIRUZKLFKWKHVXPRISUHGLFWHG
SUREDELOLW\RIWKHIXOOPRGHO H[FOXGLQJFRQVWDQW LVKLJKHUWKDQ2LOLQÀXHQFHLVWKHVXPRI
predicted probability by estimates of oil interest and vote buying, either indifferent or indifferent
plus indifferent*margin. A senator was assumed to vote yes if predicted probability was 0.5 or
KLJKHUDVLWLVFXVWRPDU\WRGHYHORSFODVVL¿FDWLRQWDEOHVIRUELQDU\RXWFRPHHFRQRPHWULFPRGHOV
Sources: See Appendix Table 1.
2QFHWKHQHJDWLYHLQÀXHQFHRIDVWDWH¶VIHGHUDOWD[HVFRQWULEXWLRQVDQG
Roosevelt’s legacy on a senators’ voting preference is considered, W1
and W2 should not have been enough to pass the UTT.
Line 5 includes the simultaneous effects on voting patterns of W1
and W2WD[FRQVLGHUDWLRQVDQGWKHRLOLQWHUHVW0RGHOVSHFL¿FDWLRQLQ
column 1 predicts that 68 senators should have voted positively, while
VSHFL¿FDWLRQVLQFROXPQ12 The predictions suggest the oil special
LQWHUHVWLQÀXHQFHGEHWZHHQDQGVHQDWRUVWRWDNHWKHFRDOLWLRQIURP
55 to 68 in column 1 and from 58 to 69 in column 2.
$QRWKHUZD\WRVHHWKHLPSRUWDQFHRIWKHRLOOREE\RQWKH¿QDOUHVXOWRI
the UTT roll call is to examine how many senators should have opposed
it on ideological grounds and how many of these senators are predicted
to have changed their position once the oil special interest is consid-
ered. Line 6 indicates the number of senators predicted to have opposed
12
2IWKHVHQDWRUVPRGHOVSHFL¿FDWLRQSUHGLFWVWRKDYHYRWHGWRUDWLI\WKHWUHDW\GLGLQ
IDFWYRWHGWRUDWLI\7KLV¿JXUHIRUVSHFL¿FDWLRQLV
:(/)$5(())(&762)7+(85587,$7+2062175($7<
621-3UR¿WV
621-¶V EHQH¿WV IURP WKH 877 UDWL¿FDWLRQ PD\ EH PHDVXUHG E\ WKH
DGGLWLRQDOSUR¿WVDWWULEXWHGWRWKHH[SORLWDWLRQRI&RORPELDQRLO7KHVH
SUR¿WVDOORWKHUWKLQJVEHLQJHTXDODUHGHULYHGIURPH[WUDFWLQJWUDQV-
SRUWLQJ DQG UH¿QLQJ &RORPELDQ RLO DQG IURP ORZHU SULFHV RQ DOO QRQ
&RORPELDQEDUUHOVRIRLOSURGXFHGDQGUH¿QHG7RHVWDEOLVKWKHPDJQLWXGH
RIDGGLWLRQDOSUR¿WVLWLVQHFHVVDU\WRNQRZWKHFRVWVWUXFWXUHDQGSULFHV
within the vertical supply chain, all indirectly owned by SONJ, but this
information is not available. An alternative is to identify a lower bound of
DGGLWLRQDOSUR¿WVDQGDVVXPHWKDW752&2¶VDQG$QGLDQ¶VSUR¿WVUHÀHFW
621-¶VDGGLWLRQDOSUR¿W,QRUGHUWRHVWLPDWHWKHVHEHQH¿WVZHFDOFXODWH
the net present value (NPV) of TROCO’s and Andian’s operations.
,GHDOO\RQHZRXOGOLNHWRNQRZWKHÀRZVRIFDSLWDOLQYHVWPHQWVDQGQHW
income for both TROCO and Andian to calculate the NPV. Conventional
VRXUFHV IRU WKLV LQIRUPDWLRQ DUH WKH DQQXDO ¿QDQFLDO UHSRUWV RI HDFK
company. These reports are available only for 1936–1938.
The information in these reports indicates that average capital stock
in 1936–1938 was $42.1 million for TROCO and $19.3 million for
In columns 1 and 2, of the three senators predicted to have voted to ratify the treaty although
13
Andian.14*LYHQWKHVKRUWSHULRGFRYHUHGE\WKH¿QDQFLDOUHSRUWVZHDOVR
examined print media, which reported that capital subscribed by TROCO
in 1920 was $39 million, presumably to invest in building production
HTXLSPHQWDQGDUH¿QLQJSODQW$QGLDQLVVXHGPLOOLRQLQERQGGHEWWR
¿QDQFHSLSHOLQHFRQVWUXFWLRQLQ±ZKLOHLWVVXEVFULEHGFDSLWDO
was $1 million. Thus, capital stock in the annual reports and the press
seem to be roughly in line.15
7KHDYDLODEOHDQQXDO¿QDQFLDOUHSRUWVLQGLFDWHDYHUDJHDQQXDOQHWHDUQ-
ings of $2 million for TROCO, $7.3 million for Andian.16 Operational net
LQFRPHÀRZVDUHFRQVWUXFWHGE\XVLQJSURGXFWLRQVWDWLVWLFVDQGD¿[HG
GROODU SUR¿W SHU EDUUHO SURGXFHG RU WUDQVSRUWHG 752&2¶V RSHUDWLRQDO
net earnings are estimated as barrels produced per year times the dollar
DPRXQWRISUR¿WSHUEDUUHOSURGXFHG7KHTXDQWLW\RIEDUUHOVSURGXFHG
and transported is calculated with production and export statistics
SXEOLVKHGE\WKH&RORPELDQJRYHUQPHQW7KHSUR¿WSHUEDUUHOSURGXFHG
LVFHQWVFDOFXODWHGDVWKHDYHUDJHSUR¿WSHU\HDU±GLYLGHG
by the average number of barrels produced. Andian’s operational net
earnings are estimated as barrels exported per year times dollar amount
RI SUR¿WV SHU EDUUHO H[SRUWHG 7KH SUR¿W SHU EDUUHO WUDQVSRUWHG LV
FHQWVWKHDYHUDJHQHWSUR¿WSHU\HDU±GLYLGHGE\WKHDYHUDJH
number of barrels transported.
TROCO’s operational net earnings are probably slightly underes-
timated. Since more than 90 percent of its production was exported,
assuming that extraction costs are stable over time, net earnings depend
on the international price. The international prices were $1.13 per barrel
in 1936–1938, slightly below the 1921–1951 period’s average of $1.25.
Andian’s net earnings are roughly accurate as transportation services
experienced stable costs during the period.17
8VLQJWKHFDSLWDOLQYHVWPHQWLQDQQXDOUHSRUWVDVDQLQLWLDO¿[HGFDSLWDO
FRVWWKHHVWLPDWHGQHWHDUQLQJVÀRZVDQGD'RZ-RQHVDYHUDJHDQQXDO
return of 5.6 percent to proxy the discount rate, the NPV for TROCO is
–$25.1 million, for Andian it is $78.2 million, and for SONJ it is $53.1
million. Even if the project had been perceived to be 40 percent more
14
,QWHUQDWLRQDO3HWUROHXP&RPSDQ\DQQXDOUHSRUWV±ZLWKLQIRUPDWLRQVSHFL¿FWR
Tropical Oil Company and National Andian Corporation. Glenbow Museum Archive (Glenbow
hereafter), Imperial Oil Collection.
15
New York Times, 14 August 1920, Oil and Gas Journal, Tulsa, 22 January 1925, 5 February
1925, 7 May 1925.
16
International Petroleum Company Annual Reports, 1936–1938 (Glenbow).
17
3UR¿W LQIRUPDWLRQ FRPHV IURP ,QWHUQDWLRQDO 3HWUROHXP &RPSDQ\ Annual Reports, 1936–
1938 (Glenbow). Barrels produced and transported come from Colombia, Ministerio de Minas y
Petroleo (1944, p. 88) and Santiago (1986, p. 63).
risky than investment in stocks (increasing the discount rate from 5.6 to 8
SHUFHQW WKHSURMHFWZRXOGVWLOOKDYHEHHQH[SHFWHGWREHSUR¿WDEOH
621-¶V 139 LQGLFDWHV WKDW UDWL¿FDWLRQ RI WKH 877 UHVXOWHG LQ DQ
LPSOLFLWVXEVLG\WKDWZDVVPDOOHUWKDQ621-¶VDFFUXHGSUR¿WV7KH86
Senate could have requested that SONJ pay voluntary taxes to compen-
sate for the reparations paid by the United States in 1921.
7KH UDWL¿FDWLRQ RI WKH 877 DOO RWKHU WKLQJV EHLQJ HTXDO IDFLOLWDWHG
SONJ crude oil production in Colombia and led to an increase in supply
and a reduction in the equilibrium price for crude oil. In turn, the price
GHFOLQHVKRXOGKDYHDOORZHG$PHULFDQUH¿QHUVWRVDYHRQFUXGHRLOLQSXW
costs.
7RFDOFXODWHKRZPXFK$PHULFDQUH¿QHUVVDYHGZHIROORZDFRQYHQ-
tional social savings approach. The intuition is that had Colombia not
exported crude petroleum to the United States, the crude price should
have been higher. How much higher depends on the price elasticity of
demand. We use a range of plausible price elasticities of demand to esti-
PDWHWKHH[SHQGLWXUHVDYHGE\UH¿QHUVFRPSDULQJDVFHQDULRZKHUHWKH
United States imported Colombia’s oil with another where it did not.
The social savings approach is a conventional tool used by economic
historians to estimate the partial equilibrium welfare effects derived from
the introduction of a product into a market (Fogel 1964; Fishlow 1965).
The main advantage of this approach is its simplicity and small data
requirements. However, the approach focuses on demand and assumes
an inelastic supply with few, if any, alternative producers entering the
market had Colombia’s crude oil not been imported into the United
States. In this case, the use of the social savings approach is appropriate
only after some simple but important institutional considerations are
included in the set-up of the exercise.
)LUVWWKH86FUXGHRLOPDUNHWZDVVHJPHQWHGEHWZHHQWKH3DFL¿FDQG
the eastern and central markets. Crude oil production was located in large
quantities in only a few states (Pennsylvania and West Virginia in the east;
Louisiana, Texas, Oklahoma, Indiana, and Illinois in the central region;
and California in the west) and this created regionally separate markets.
By 1920, railroads and pipelines provided the necessary transport infra-
structure to facilitate intermediation between regional market segments
in the east and the central United States. And once the Panama Canal
ZDVIXOO\RSHQLQFUXGHRLOWUDGHEHWZHHQWKH3DFL¿FFRDVWDQGWKH
eastern and central United States became easier. Price convergence was
DFKLHYHG LQ WKH V %DLQ 0DXUHU DQG <X SS ±
Libecap 1989, p. 835). Thus, we can think of the American crude oil
market as a segmented market for most of the period of analysis, with
&RORPELDQRLOÀRZLQJLQWRWKHHDVWHUQDQGFHQWUDOVHJPHQWRIWKHPDUNHW
6HFRQG LQ WKH V ZKHQ WKH 877 ZDV UDWL¿HG WKH HDVWHUQ DQG
central crude oil market was not only integrated but production operated
in a highly competitive manner. Independently of whether a producer
was part of a major oil company or an independent, both extracted oil
IURPDVSHFL¿FRLO¿HOGDVIDVWDVSRVVLEOHWRUHGXFHWKHQHJDWLYHHIIHFWV
of competition when exploiting a common pool resource. This market
structure dominated the crude oil market until 1933, when it became a
cartel (Libecap 1989). Once the crude oil cartel began effective opera-
tion, import controls were set up in the United States. In turn, the share
of Colombian crude oil exports to the United States declined from an
average of 72 percent in 1926–1933 to 26 percent in 1934 and never
recovered. Thus, we focus our analysis on the social savings up to 1933.
The calculation of social savings assumes a range of price elasticity of
demand between relatively inelastic 0.5 and relatively elastic 5. The U.S.
mean imports of Colombia’s crude oil over 1926–1933 were 11.9 million
barrels per annum (see Table 4 line 1). The eastern and central U.S. mean
crude oil consumption was 699 million barrels per annum. Colombia’s
oil mean market share was 1.72 percent. The Oklahoma crude oil price
was used by the U.S. Bureau of Mines as the reference for the eastern
and central U.S. market segment (U.S. Bureau of Mines, various years).
The mean real Oklahoma price was $1.52 per barrel. The mean price
increase had Colombia’s oil not reached the eastern and central U.S.
market segment ranges from 0.055 to 0.006 dollars per barrel, which
implies that crude oil consumers in this market saved an annual mean
expense ranging from $37 million per annum to $3.7 million. The NPV
of total social savings, 1926–1933, ranges from $243 million if the price
elasticity of demand was 0.5, to $24 million if the price elasticity was
5.0 (see line 8). If the price elasticity of demand was lower than 4.9, the
social savings should have been higher than the $25 million in reparation
WKH8QLWHG6WDWHVSDLG&RORPELDDIWHUUDWL¿FDWLRQRIWKH877
To examine how sensitive this result is, the West Texas price, that
was about one-third lower, is used instead. A lower crude price renders
smaller social savings, ranging from $149 to $14 million (see line 9). If
the elasticity was lower than 2.9, reparations should have been smaller
than social savings. In sum, as long as the price elasticity of demand was
lower than about 3, the total social savings of eastern and central U.S.
UH¿QHUVIURPWKHÀRZRI&RORPELDQRLOVKRXOGKDYHEHHQKLJKHUWKDQWKH
276
TABLE 4
SOCIAL SAVINGS FOR THE UNITED STATES RESULTING FROM UTT
Price Elasticity of Demand 0.5 1.0 2.0 3.0 4.0 5.0
(1) U.S. crude oil imports from Colombia (million barrels per annum) (CM) 11.9 11.9 11.9 11.9 11.9 11.9
(2) East and central U.S. crude oil consumption (million barrels per annum) (CC) 699 699 699 699 699 699
(3) Share of Colombia imports in east and central U.S. consumption (percent) (CM/CC) 1.72 1.72 1.72 1.72 1.72 1.72
UHSDUDWLRQV WKH 8QLWHG 6WDWHV SDLG WR &RORPELD DIWHU UDWL¿FDWLRQ RI WKH
UTT.
Finally, the distribution of social savings within America is an impor-
tant issue. We compare crude oil and oil derivatives prices and provide
a rough indication of the distribution of the gains to the United States
derived from the effects of the treaty.
7KHRLOUH¿QLQJLQGXVWU\ZDVDFRQFHQWUDWHGROLJRSRO\RSHUDWHGE\WKH
oil majors. Thus, it is not possible to use only the market price of crude
RLOIXHORLODQGJDVROLQHWRHVWLPDWHWKHSUHFLVHGLVWULEXWLRQRIEHQH¿WV
Cost, productivity series, and behavioural assumptions, in addition to
price series, are required. This information is not available.
+RZHYHU ZH FDQ GHYHORS D UH¿QLQJ GHULYDWLYHV DJJUHJDWH SULFH
index that we can compare to the Oklahoma crude price index and John
Kendrick’s sectoral productivity statistics and infer a preliminary view on
the distribution of social savings. Figure 1 presents an output weighted
SULFH LQGH[ IRU UH¿QLQJ GHULYDWLYHV FRPSRVHG RI JDVROLQH
fuel oil, and kerosene and an index for the Oklahoma crude oil price.
7KUHHPDLQWUHQGVDUHZRUWKKLJKOLJKWLQJ)LUVWWKH¿JXUHLQGLFDWHVWKDW
crude oil price plunged between 1926 and 1928. Colombia’s petroleum
contributed to this decline, with a market share of more than 2 percent in
the eastern and central U.S. crude oil market segment. The price stabi-
lized at about 50 percent of the 1926 value. The derivatives price index
during the same period also declined, but only to 80 percent of the 1926
OHYHO 7KH UH¿QLQJ LQGXVWU\ FDSWXUHG PRVW RI WKH EHQH¿WV WKH 8QLWHG
States derived from the price decline induced by Colombia’s crude oil.
Second, for the relevant social savings period, 1926–1933, the average
crude oil price index was 56 and derivatives price index was 98. Thus, oil
UH¿QHUVVDYHGDSSUR[LPDWHO\RQHKDOIRIWKHLUFUXGHRLOH[SHQVHVDQGGLG
not pass along any of these savings to oil derivatives consumers, main-
taining the consumer surplus unchanged.
7KLUGWKH¿JXUHVKRZVWKDWDIWHUSULFHVWHQGWRGHFOLQHVORZO\
IRUERWKFUXGHRLODQGUH¿QLQJGHULYDWLYHV$W¿UVWJOLPSVHRQHPD\LQIHU
WKDW ¿QDO FRQVXPHUV EHQH¿WHG IURP GHFOLQLQJ SULFHV IRU FUXGH RLO DQG
UH¿QHGSURGXFWV%XWSURGXFWLYLW\ZDVJURZLQJIDVWHUWKDQRXWSXWSULFHV
were declining. Between 1929 and 1937 crude oil production experienced
DQSHUFHQWWRWDOIDFWRUSURGXFWLYLW\JURZWKZKLOHWKHUH¿QLQJLQGXVWU\
experienced a 25 percent total factor productivity growth (Kendrick
1961, pp. 400, 448). Productivity data suggests that the producer surplus
RIUH¿QHUVJUHZEHFDXVHFUXGHRLOSULFHVGURSSHGDQGSURGXFWLYLW\JUHZ
while for crude oil suppliers the producer surplus grew because produc-
tivity was growing faster than the crude oil price declining.
120
100
80
60
40
20
0
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
Oklahoma crude oil Oil refining derivates
FIGURE 1
INDEX NUMBERS FOR CRUDE OIL AND OIL REFINING DERIVATIVES PRICES
(1926=100)
CONCLUSION
Appendix
APPENDIX TABLE 2
DESCRIPTIVE STATISTICS
Variable Name NObs Mean Standard Deviation Minimum Maximum
Ratify 88 0.7841 0.4138 0.00 1.00
Reduce 88 0.2386 0.4287 0.00 1.00
Peace with Germany 88 0.7159 0.4536 0.00 1.00
Crude 88 0.0109 0.0297 0.00 0.18
5H¿QLQJ 88 0.0115 0.0227 0.00 0.13
Oil interest 88 0.0224 0.0493 0.00 0.24
W1 88 0.1761 0.7477 –1.00 0.99
W2 88 –0.0645 0.4404 –1.00 0.88
Tax ratio 88 44.5000 25.5428 1.50 87.50
Indifferent 88 0.5000 0.5028 0.00 1.00
Margin 88 0.2760 0.3142 0.00 1.00
Indifferent*Margin 88 0.1318 0.2589 0.00 1.00
Source6HHWH[WDQG$SSHQGL[7DEOHIRUVRXUFHVDQGGH¿QLWLRQVRIYDULDEOHV
REFERENCES
Alesina, Alberto, and David Dollar. “Who Gives Foreign Aid to Whom and Why?”
Journal of Economic Growth 5, no. 1 (2000): 33–63.
American Petroleum Institute. Petroleum Facts and Figures 1HZ <RUN $PHULFDQ
Petroleum Institute, various years.
Bain, Joe. 7KH (FRQRPLFV RI WKH 3DFL¿F &RDVW 3HWUROHXP ,QGXVWU\ 3DUW ,, 3ULFH
%HKDYLRXU$QG&RPSHWLWLRQ Berkeley: University of California, 1945.
Ball, Richard, and Christopher Johnson. “Political, Economic, and Humanitarian
Motivations for PL 480 Food Aid: Evidence from Africa.” Economic Development
and Cultural Change 44, no. 3 (1996): 515–37.
Bell, Purl Lord. Colombia: A Commercial and Industrial Handbook. Washington, DC:
GPO, 1921.
Berger, Daniel, William Easterly, Nathan Nunn, and Shanker Satyanath. “Commercial
,PSHULDOLVP" 3ROLWLFDO ,QÀXHQFH DQG 7UDGH GXULQJ WKH &ROG :DU´ American
Economic Review 103, no. 2 (2013): 863–96.
Bordo, Michael, and John Landon-Lane. “Exits from Recessions: The U.S. Experience
1920–2007.” NBER Working Paper No. 15731, Cambridge, MA, February, 2010.
Bucheli, Marcelo. “Negotiating under the Monroe doctrine: Weetman Pearson and the
Origins of U.S. Control of Colombian Oil.” %XVLQHVV +LVWRU\ 5HYLHZ 82, no. 4
(2008): 529–53.
Bushnell, David. Colombia: Una Nación a Pesar de Si Misma. Bogotá: Editorial
Planeta, 2007.
Coatsworth, John. “Liberalism and Big Sticks: The Politics of U.S. Interventions in
Latin America, 1898–2004.” Paper Presented at the Conference “Liberalism and
Its Legacies: A Symposium of Latin American History in Honor of Charles A.
Hale,” University of Iowa, Iowa City, Iowa, 4 March 2006.
Colombia. Cámara de Representantes. Informe de la Comisión Investigadora y
Documentos Relacionados con el Contrato del Oleoducto entre el Gobierno